šŸ¤

Prisoner's Dilemma: Cooperation vs Defection

Nov 9, 2025

Overview

The transcript explains the Prisoner’s Dilemma, shows payoffs and dominant strategies, gives a marketing analogue, and discusses rational agents versus real human behavior.

Prisoner’s Dilemma Setup

  • Two suspects (Mr. Blue, Ms. Red) arrested for a minor crime; serious crime lacks evidence.
  • Police separate them; each chooses to Stay Silent (cooperate) or Betray (defect).
  • Outcomes depend on combined choices; incentives push toward defection.

Payoff Structure and Strategies

  • Mutual silence: each gets 1 year in prison; best total outcome for the pair.
  • One betrays, other silent: betrayer goes free; silent gets 3 years.
  • Mutual betrayal: each gets 2 years; worse than mutual cooperation.
  • Individual reasoning: defection dominates regardless of the partner’s action.

Prisoner’s Dilemma Payoff Table

Blue \ RedStay SilentBetray
Stay SilentBlue: 1 year; Red: 1 yearBlue: 3 years; Red: 0 years
BetrayBlue: 0 years; Red: 3 yearsBlue: 2 years; Red: 2 years

Group vs Individual Outcomes

  • Group optimum: both cooperate (total 2 years).
  • Individual incentive: always defect to avoid worst-case and seek best-case.
  • Result: both defect; total and individual outcomes worse than mutual cooperation.

Marketing Analogue (Advertising Game)

  • Two identical cigarette firms: Red Strikes and Smooth Blue.
  • Choices: Advertise a lot or Not Advertise; market of 100 smokers; price $2; ad cost $30.
  • No advertising: random split; 50/50 buyers; each earns $100.
  • One advertises: advertiser gets 80 buyers ($160 āˆ’ $30 = $130); other gets 20 buyers ($40).
  • Both advertise: split 50/50; each earns $100 āˆ’ $30 = $70.
  • Firms can talk, but lack enforceable obligations; incentive to advertise remains.

Advertising Game Payoff Table

Firm \ FirmDo Not AdvertiseAdvertise
Do Not AdvertiseEach: $100Advertiser: $130; Non-ad: $40
AdvertiseAdvertiser: $130; Non-ad: $40Each: $70

One-Shot vs Repeated Interactions

  • One-shot assumption: single play; no relationship or punishment possible.
  • Repeated play changes incentives: allows reputation, cooperation, and punishment.
  • Model simplification: players treated as predictable, rule-following agents.

Rational Agents vs Real People

  • Rational agent: always chooses the option predicted best for self; ignores others’ gains.
  • Real behavior study: 40 people, one-off computer games, no communication.
  • Observed cooperation average: 22%; some never cooperated; some always did; many in between.
  • Implication: humans are social; sometimes adopt group perspective even without obligations.
  • Model value: highlights dilemmas where self-interest can harm both self and group.

Key Terms & Definitions

  • Prisoner’s Dilemma: game where individual rationality leads to worse joint outcomes.
  • Cooperate (Stay Silent): choose group-beneficial option risking exploitation.
  • Defect (Betray): choose self-beneficial option regardless of partner’s choice.
  • Dominant Strategy: best action regardless of the opponent’s move (defection here).
  • Rational Agent: hypothetical decision-maker maximizing own payoff consistently.
  • One-shot Game: single interaction with no future consequences.
  • Payoff Matrix: table listing outcomes for all action combinations.

Action Items / Next Steps

  • Recognize settings with Prisoner’s Dilemma structure in real decisions.
  • Consider repeated interactions or enforceable commitments to support cooperation.
  • Use rational-agent models to diagnose incentive problems; adjust rules to align interests.